tiny.x offset matrix rows not orthonormal

tiny.x offset matrix rows not orthonormal

Post by Cronus3 » Sun, 06 Jul 2003 03:19:45

After some investigating I found that my final bone matrices were bad (not
rigid body transformations).  I found that the offset matrices' upper 3x3
rotation part rows were not orthonormal, and when these were multiplied
through errors accumulated leaving me with a very distorted matrix.

Now, AFAIK the offset matrix transforms the joint relative to the character
to the origin.  This should be a rigid body transformation, should it not?

Why then, is there scaling in the offset matrx?  For example, in tiny.x

SkinWeights {


Take the top row, row0: ||row0|| = 1.27.

Now, there is not much that can be done about the data contents as they are,
but how does the SDK D3DX animation functions handle this?  Obviously, the
SDK samples animate tiny.x correctly.  Are they normalizing the rows

If anyone could give some insight I'd appreciate it very much.


1. Orthonormal matrix algorithms


I am looking for references concerning various algorithms/math theory
relative to orthonormal matrices.

I am currently working on surface matching algorithms in which we use
minimisation algorithms to calculate transformation matrices. Up to now,
I have not found specialised litterature on orthonormal matrices calculations.
I have been told group theories address these issues, but have been
unlucky as to references so far.

Any help/hints would really be appreciated


Louis Amiot

2. particles question #x

3. Is modelview matrix orthonormal?

4. How Do I Capture An Image???

5. Row major or column major matrices? - SUMMARY POSTING

6. Adding a glow effect to the white areas in a map?

7. Row major or column major matrices?

8. A Newbie FAQ... in the works

9. transformation matrix - translation in end row or column?

10. Projection matrix for offset eye point

11. bone offset matrices

12. SkinWeights and Offset Matrix in .X File

13. REQ: help with Offset filter, (not in PS directory)